Pavlov 0.2.0

A modular reinforcement learning pipeline with Keras

Pavlov is a modular, composable reinforcement learning API, with models written in Keras with a Tensorflow backend.

Use Pavlov if you want a reinforcement learning library that:

• Allows for easy and fast composition of various algorithms, exploration strategies, and state preprocessors

• Allows for arbitrarily defined custom network architectures

• Works with any environment that follows the OpenAI Gym interface

• Comes with its own Docker image for both GPU and CPU usage

Pavlov is distributed under the MIT license.